Foxg1 Organizes Cephalic Ectoderm to Repress Mandibular Fate, Regulate Apoptosis, Generate Choanae, Elaborate the Auxiliary Eye and Pattern the Upper Jaw

Gnathostome jaw patterning involves focal instructive signals from the embryonic surface cephalic ectoderm (SCE) to a fungible population of cranial neural crest. The spatial refinement of these signals, particularly for those patterning the upper jaws, is not fully understood. We demonstrate that Foxg1, broadly expressed in the SCE overlying the upper jaw primordia, is required for both neurocranial and viscerocranial development, including the sensory capsules, neurocranial base, middle ear, and upper jaws. Foxg1 controls upper jaw molecular identity and morphologic development by actively inhibiting the inappropriate acquisition of lower jaw molecular identity within the upper jaw primordia, and is necessary for the appropriate elaboration of the λ-junction, choanae, palate, vibrissae, rhinarium, upper lip and auxiliary eye. It regulates intra-epithelial cellular organization, gene expression, and the topography of apoptosis within the SCE. Foxg1 integrates forebrain and skull development and genetically interacts with Dlx5 to establish a single, rostral cranial midline.


Introduction
The evolutionary success of vertebrates is intimately tied to developmental innovations affecting the head, including those leading to the elaboration of the brain, the emergence of the cranial neural crest (CNC) and ectodermal placodes, and the manifestation of jaws (1)(2)(3). As the vast majority of vertebrates are gnathostomes, and thus possess jaws, the true enormity of the observed radiation and diversification of vertebrates must be regarded in the particular light of the acquisition and modification of jaws . Jaws are prehensile oral apparatuses that can be further defined by their possession of two appositional units -the upper and lower arcades -that are articulated (hinged) and which largely arise during development from the first branchial arch (BA1) (5-12, 23-31).
Both the jaws and their BA1 primordia are characterized by their polarity: Within BA1, polarity is bidirectional and is roughly centered midway along the arch, extending from the so called 'hinge' proximally through the maxillary BA1 (mxBA1) and distally through the mandibular BA1 (mdBA1) (Figure 1 and Suppl. Figure 1; 5-9, 30-32). This directionality is subsequently mirrored by the associated structural polarity, centered at the functional jaw articulation, of the realized upper and lower jaw arcades. The fundamentality of this particular positional directionality for jaw development was initially molecularly revealed by the induced morphologic homeotic transformation of lower jaws into hinge-centric, mirror-imaged upper jaws in transgenic mice lacking a linked pair of Distal-less genes (Dlx5/6) whose expression is restricted to the lower jaw primordia (i.e., mdBA1) (6, 32, 33). Importantly, jaws exhibit additional levels of polarity: for instance, the upper jaws are overall more intimate with the neurocranium (the portion of the skull that houses the brain and primary sensory capsules) and its development, while the lower jaws are in closer developmental association with the emerging heart (5)(6)(7)(8)(9)(10)(11)27). A functional manifestation of this additional polarity is the evolutionary inclusion of a premaxillary component derived from the olfactory placode-associated frontonasal prominences (FNP) in the upper jaw arcades of gnathostomes (except chondrychthyans; 4-31).
The developmental system that establishes these polarities and patterns the developing jaws involves an intricate spatiotemporal series of reciprocal inductive and responsive interactions between the cephalic epithelia (both ectodermal and endodermal) and the subjacent CNC mesenchyme (6-10, 25, 30-50; reviewed in 9). Notably, augmentations of the developmental mechanisms and dynamics specifically regulating the cephalic epithelia and mesenchyme of the nascent upper jaw primordia (i.e., mxBA1 and FNP) and their associated structures have been connected with a number of significant gnathostome evolutionary events (8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(51)(52)(53)(54)(55)(56)(57)(58)(59)(60)(61)(62)(63), including: the teleostean radiation (impelled in part by making the maxillary-premaxillaryneurocranial connectivity less rigid and more mobile), the colonization of land by tetrapods (being in part enabled by acquisition of internal choanae thereby enabling respiration with a closed mouth), both the mammalian ability to masticate while breathing (enabled by the presence of a secondary palate) and the possession in mammals of highly expanded and coordinated olfaction and respiration (propelled by the evolution and elaboration of often labyrinthine turbinal bones in the nasal cavity and paranasal sinuses in the associated cranial skeleton), the cetacean radiation, as well as the primary split of primates into strepsirhini or haplorhini (following the truncation of the nasal cavity, the loss of the rhinarium [the specialized glabrous skin these components of the Foxg1 -/-neurocranium ( Figure 2 and Supplementary Figure 5). The absence of an expanded telencephalon and the presence of an abnormal optic stock (see below) left the dermatocranial frontal bones clearly highly dysmorphic, reduced medio-laterally, and lacking orbital laminae. Foxg1 -/-neonatal calvaria also exhibited open anterior fontanelles between the frontal bones, displaying distinctly pyramidal shaped extended patency of the posterior interfrontal (metopic) sutures ( Figure 2H). The parietals and interparietals, moreover, were often smaller which led to wider lambdoidal sutures.
Within the neurocranium, the otic capsule (OTC) envelopes the vestibular and cochlear apparatuses in the pars canalicularis and pars cochlearis, respectively; both of these were dysmorphic in Foxg1 -/-neonates ( Figures 2I-J'). Reflective of defects of the semi-circular canals, within the pars canalicularis both the prominentia semicircularis anterior and posterior were abbreviated. Moreover, the oval and round windows appeared anomalous in size and orientation in the capsules. Within the pars cochlearis, the cochlea failed to make the usual one and a half turns. Additionally, two extra-capsular structures that are normally synchondrotic with the otic capsule, the tegmen tympani (TT; a mammalian neomorphic structure) and the styloid process, were both found to be dysmorphic and disconnected from the OTC and were otherwise found anomalously associated with the altered middle ear skeleton (see below). A third extra-capsular structure that normally connects the rostromedial margin of the pars cochlearis to the ala temporalis of the alisphenoid, the alicochlear commissure, was also dysmorphic and misrouted from joining the OTC to joining with the remnants of the styloid process (see below).
While relatively small, the murine OPC is comprised of pre-optic and post-optic pillars that, together with the laterally placed ala orbitalis and medial presphenoid, frame the optic foramen. Normally, the anteriolateral margin of the ala orbitalis rostrally extends a sphenethmoidal (orbitonasal) commissure that meets the dorsolateral NC, thus forming the lateral boundary of the orbitonasal fissure that separates the nasal and optic capsules. Likewise, the ala orbitalis normally extends a caudal posterior commissure to meet the taenia marginalis of the neurocranial sidewall. The OPC of Foxg1 -/-neonates, however, lacked a pre-optic pillar and most of the ala orbitalis ( Figures 2K-P), and thus also any connections to the NC. Instead, a single, aberrantly straight rod-like post-optic pillar with a precocious ossification center, a sharp thin posterior commissure directed towards a diminished taenia marginalis, and a precociously ossified ala hypochiasmatica was found extending laterad from the caudal boundaries of the presphenoid. The preshenoid was found to be dysmorphic and evincing extended, precocious lateral ossification but a distinct lack of ossified unity at (and across) the midline.
Development and morphogenesis of the NC was also significantly perturbed in Foxg1 -/-mutants (Figures 2F, K-R; Supplementary Figure 4). Because the normally developing murine nasal cavity is structurally complex and developmentally intimately tied to the λ-junction and the forming upper jaws a broader description of its normal skeletal development is expedient. The perinatal skeleton of the murine nasal cavity (cavum nasi) is composed of two bilateral NC separated medially by an extension, the nasal septum (septum nasi), of the midline trabecular basal plate (TBP) that underlies the brain rostral to the hypophysis. The TBP is formed from paired midline extensions (trabeculae cranii) running rostrad from the center of the basi-sphenoid (at the hypophysis), through the presphenoid and extending rostrad to form the septum nasi between the NC. These paired structures typically condense and chondrify in such a manner that a single midline cartilaginous structure and presphenoidal ossification center is normally seen in skeletal preparations. To either side of the rostroventral part of the nasal septum lie paraseptal cartilages that house the vomeronasal organs (VNO). The NC themselves are paired cartilaginous sacs formed from frontonasal CNC condensing around the invaginating olfactory pits, each generating the roof, sidewall and floor of the " 8 cavum nasi.
Anteriorly, the dorsal NC normally arcs medially to fuse to the nasal septum thereby forming the midline grooved anterior roof, the tectum nasi. The anterior extremity of the tectum extends beyond the overlying dermatocranial nasal bones and ends with the cupula nasi anterior and the outer nasal cartilages that frame the external nares and support the rhinarium. The caudal border of the tectum, the crista cribroethmoidalis, forms a sharp crest that meets the obliquely oriented lamina cribrosa that forms the caudal NC roof, underlies the olfactory bulbs, and is perforated by the olfactory nerves and ethmoidal vessels.
Each NC sidewall, or paries nasi, is grossly composed of three subdivisions: the pars anterior, pars intermediale, and pars posterior. It is the paries nasi that essentially establishes the functional organization of the interior nasal spaces. Rostrally, the elongate, narrow pars anterior helps to establish the anterior vestibule The internal landscape of each NC is ontogenetically dominated by the progressive development and ramification of turbinals (turbinates, or nasal conchae), capsular projections into the nasal fossa from the tectum, paries nasi and lamina transversalis that greatly increase the surface area covered by the various nasal epithelia. The anterior-most murine turbinal, the atrioturbinal, develops from the lamina transversalis anterior in each nasal fossa; caudally, it is followed by a nasoturbinal descending from the tectum and a maxilloturbinal that projects from the pars intermediale. Both of these latter turbinals normally extend rostrally through much of the pars anterior. A cartilaginous turbinal crest, the crista semicircularis, projects inward from the prominentia lateralis thereby segregating the anterior and middle nasal chambers. Within the prominentia lateralis, a horizontal lamina functionally offsets an upper recessus frontoturbinalis and a lower recessus maxillaris. From the pars posterior and lamina cribrosa arise multiple frontoturbinals (superiorly) and several ventral ethmoturbinals. Most of the above described complex NC structures were found to be deficient or lacking in the absence of Foxg1. Overall, the Foxg1 -/-perinatal snout was compressed mediolaterally, dorsoventrally and rostrocaudally, and this is reflected in changes to the NC (Figure 2 and Supplementary Figure 5). A nascent rhinarium and external capsular narial openings were patent in mutant perinates (see Fig. 9C3); however, the cupula nasi anterior was smaller and the outer nasal cartilages were anomalous in size and orientation. The " 9 initial intranarial epithelium was thicker in Foxg1 -/-mutants, however, and led to a smaller vestibular entrance.
The tectum nasi with shortened nasoturbinals, the nasal septum, and the lamina transversalis anterior with atrioturbinals were each present in mutant skulls though they were all generally less refined and thicker. Pars anterior with diminished maxilloturbinals also characterized the NC in Foxg1 -/-perinates. However, caudal to the laminae transversalis anterior and atrioturbinals, the floor of the mutant NC was disorganized and asymmetric; and, while present, paraseptal cartilages were dysmorphic, misaligned and lacked VNO.
Although affected by the loss of Foxg1, the rostral NC was of more or less patent; the middle and caudal NC, however, were devastated by either the truncation or the complete loss of structure. The pars intermediale lacked both an expansive prominentia lateralis and its associated crista semicircularis; thus, there were no recessus laterales nor sulci anteriolaterales. Presumptive processus maxillae posterior did, however surprisingly, project from the diminished pars intermediale of mutant skulls. In this mid region of the NC, the septum of Foxg1 -/-mutants typically became discontinuous and asymmetric dorsoventrally. The tectum nasi of mutant perinates ended at an abrupt crista cribroethmoidalis. Significantly, however, caudal to the crista there were no laminae cribrosae, and the caudal NC was represented only by ventrolateral, unadorned cartilaginous plates. Thus, there were no cupulae posterior, frontoturbinals or ethmoturbinals nor any of their support structures. The caudal nasal septum, moreover, presented no dorsal extension and was represented only by a cylindrical interorbitaI septum which was itself typically abnormally disconnected from the remainder of the TBP just rostral to the presphenoid. Notably, the loss of the posterior NC meant that the nasal cavities of Foxg1 -/-mutants did not empty ventrally into a nasopharyngeal duct, and they would have but for a thin caudal epithelial barrier emptied directly into the cavum cranii itself.

Upper Jaw Dermatocranial Defects of Foxg1 -/-mutants. A number of dermatocranial elements, including
jaw elements, normally develop in association with the NC, OPC, and TBP (5,7,11,135,136). For instance, paired nasal bones overlap the tectum nasi while frontal bones dorsally cover the olfactory bulbs, overlap the crista cribroethmoidalis, and are associated with the OPC. Laterally and ventrally, the maxillae cover the pars intermediale and pars posterior, while lacrimal bones are associated with the maxillae and orbital face of the NC. The lateral and ventral pars anterior are covered by the premaxillae, the palatine (palatal) processes of which underlie the paraseptal cartilages. Caudal to this, the paired vomers frame the ventral nasal septum while the maxillary and palatine bones overlap the caudal NC and TBP (including the presphenoid), respectively.
The defects associated with the tectum nasi and pars anterior of the NC were accompanied by alterations of the nasal and premaxillary bones ( Figure 2). The nasal bones of Foxg1 -/-mutants appeared mediolaterally compressed, often asymmetric, and rostrally under ossified. Premaxillae were associated with the lateral and ventral aspects of the anterior NC in Foxg1 -/-mutants; however, the inter-premaxillary foramen was pushed rostrad and extended inter-premaxillary sutures were evinced. The bodies of the premaxillae contained incisors, though they were slightly anomalous in shape. Oddly, the rostral tips of each incisor was typically associated with an extremely small, yet very clear, independent mineralized nodule (Supplementary Figure 5). Moreover, two of 15 premaxillae specifically examined housed ectopic incisors. The maxillary processes were shortened and there were no discrete palatine processes per se projecting caudally at the midline and overlying the dysmorphic paraseptal cartilages.
The maxillae are viscerocranial bones (i.e., BA-derived; see below) and are the largest elements of the perinatal upper jaw. They mature to take a significant part in forming the mature nasal, orbital and oral cavities and are thus discussed here in this context. In Foxg1 -/-neonates the maxillae were dysmorphic, in " 10 part because there were no palatine (palatal) processes as such projecting rostrad at the midline and framing the fenestra basalis. Rather, in the mutant maxillae, un-elevated (see below) and asymmetric osseous plates of the maxillary body were seen projecting both toward the midline (though they were never close to abutting and thus left an unossified cleft) and rostrad toward the premaxillae (though never reaching them). Moreover, between these plates, small isolated ectopic ossifications were encountered. Loss of the prominentia lateralis of the NC was associated with the loss of the frontal process of the maxillae and, while projecting caudally, the maxillary zygomatic processes showed less of an arc and were thinner. The bodies of the maxillary bones of Foxg1 -/-mutants housed developing molars. The lacrimal bones, usually associated with the maxillary frontal processes and the prominentia lateralis, were smaller and misshaped. Paired medial vomer bones, which usually underlie the maxillary palatine processes and are associated with the nasal septum, were present but did not elevate and were often asymmetric (following asymmetries associated with the septum nasi).
Overall, the junction of the primary and secondary palates (such as they were) was dysmorphic, and the fenestra basalis (the incisive foramen precursor and a very prominent attribute of palatal development in mice) is mostly obscured in Foxg1 -/-mutant perinates by the aberrant medial ossification patterns. Moreover, patency of the maxillary-premaxillary sutures associated with the elemental bodies of these bones was often disrupted by aberrant synostosis; synostosis was also found across the palatine-maxillary sutures. The palatal shelves of the palatine bones did project toward the midline. Notably, however, none of the 'palatal' shelves of the premaxillae, maxillae or palatine (here, the lamina horizontalis) bones were properly elevated from the ventral neurocranial base of either the NC or the TBP, and thus mutants lacked any suprapalatal respiratory passage.
Importantly, the upper jaw arcade also includes the squamosal, which is a key integrating dermatocranial element contributing to the posterior orbit, the cranial (temporal) sidewall, and the functional jaw articulation. The normal squamosal is characterized by a small central body that houses the mandibular (glenoid) cavity and from which a number of lamina and processes extend. A rostrodorsal squamosal lamina, overlying the parietal plate of the chondrocranial sidewall, extends rostrad to the frontal bone and the lamina obturans of the alisphenoid, while a zygomatic process arcs out to the jugal in the zygomatic (orbital) arch, and a ventrocaudal sphenotic lamina runs toward the cupula cochlea and the sphenoid. Moreover, there are additionally two posterior processes on the murine squamosal: a retrotympanic process that runs caudad to overlie the TT and middle ear elements and a caudal process that extends dorsocaudally between the retrotympanic and the squamosal lamina. Most of these components of the squamosal bones of the Foxg1 -/skulls exhibited deficiencies. While an articulation with the condylar process of the dentary was present, each lamina and process of the squamosal bones of Foxg1 -/-mutants was either truncated (e.g., squamosal, sphenotic, caudal and zygomatic) or absent (e.g., retrotympanic).

mutants.
Viscerocranial elements originate from the BA, consist of both splanchnocranial (chondrocranial) and dermatocranial elements, and include the jaws and middle ear ossicles (5,7,11,135,136). For instance, from each mxBA1 are derived the incus (splanchocranial), the maxillae, palatine, pterygoid, squamosal, jugal (dermatocranial) and the alisphenoid (both). From mdBA1 arises the core of the lower jaw viscerocraniumthe chondrocranial Meckel's cartilage (MC; the proximal end of which yields the malleus) -and the dermatocranial dentary, ectotympanic, and gonial bones that are associated with the MC, the malleus and the auditory bulla. The second, or hyoid, arch (BA2) viscerocranium lacks a dermatocranial contribution in mice; its splanchnocranial contribution, however, consists of the various derivative components of Reichert's " 11 cartilage. These include the stapes of the middle ear, the styloid process (itself formed of a proximal tympanohyal attached synchondroticly to the otic capsule at the crista parotica and a curved elongate stylohyal covering the round window), as well as the lesser horns and part of the body of the hyoid bone.
In Foxg1 -/-mutants, the development of the largest viscerocranial elements of the lower jaws, the BA1-derived dentary and body plus rostral process of MC, were largely spared; the remainder of the BA1 and proximal BA2 splanchnocranial elements were, however, not ( Figure 2). The malleus and incus of Foxg1 -/mutants failed to form a synovial joint; instead, they were synchondrotic between the neck of the malleus and the body of the incus. The crus brevis of the incus was itself abnormally synchondrotic with an enlarged dorsal ectopic cartilage that, based on its position overlying the incus and rostrocaudal to the pars canalicularis, we have taken as the remnant of TT which has been detached from the otic capsule (labelled as 'in:tgt:ma:ect' in Figure 2J'). The mxBA1-derived alisphenoid, including the dermatocranial lamina obturans and the ala temporalis that forms the cartilaginous base of the alisphenoid and underlies the trigeminal ganglion, was found to be significantly altered in Foxg1 -/-mutants. For instance, the horizontal lamina was thinner and both the prominent pterygoid process and the anterolateral process were lacking. Moreover, the mutant lamina typically was not met by an alicochlear commissure coming from the pars cochlearis; instead, a mediolaterally oriented ectopic cartilaginous strut was present in the region (labelled as 'sty:ect:alt' in Figures 2J', 2N). In the mutant skulls, this strut was seen to first run laterally toward the malleus before bending caudally and becoming synchondrotic with the stylohyal of the styloid process. Rather than a tympanohyal attaching to the crista parotica, Foxg1 -/-mutants topographically presented a small cartilage devoid any skeletal connection to the otic capsule. Furthermore, the stapes was only loosely associated with the window and lacked both a stapedial foramen and any functional connection with the crus longus of the incus.
The dermatocranial elements of the middle ear and nasopharynx were likewise found to be affected by the loss of Foxg1. In addition to above mentioned maxillary, palatine and squamosal bones, Foxg1 -/mutant mxBA1-associated pterygoid bones were dysmorphic. The pterygoids, normally developing in close association with the lateral margin of the basisphenoid and occupying the lateral margins of the nasopharynx, were malformed, dorsoventrally flattened and synostotic with the basisphenoid. Two proximal (i.e., hinge regional) mdBA1 associated dermal bones, the gonial and ectotympanic, were also slightly aberrant in Foxg1 -/-skulls. The gonial, which normally contributes to the middle ear by its direct osseous investment of the cartilaginous malleus, was shortened, and the breadth of the curvature of the mutant ectotympanic bone, which normally rims the tympanic and contributes to the auditory bulla, was found to be condensed in the mutant skulls, consistent with changes in the meatus.
Phenotypic analysis of Foxg1 -/-crania from E15.5 to P0 demonstrates that Foxg1 regulates the development of the skeleton associated with the λ-junction, including the upper jaws, NC, and OPC, and their associated support skeletons. Furthermore, the loss of Foxg1 has additional significant consequences for the pattern and morphogenesis of much of the cranial skeleton, including each primary sensory capsule, the middle ear, and their associated structures. The skeletal defects of the perinatal mutant skull were also accompanied by disruptions of many of the associated soft tissues (Supplementary Figures 3, 4). For instance, in addition to the absence of the VNO itself, development of the ductal systems of both the orbital (e.g., the harderian glands) and intra-NC glands were vitiated. Moreover, the presence, topography, and transitions of the extra-capsular, respiratory, transitional and olfactory epithelia were anomalous in Foxg1 -/mutants. Though present, both the hypophysis and trigeminal ganglion additionally presented anomalies of development, as notably did the eyes, the eyelids and the aforementioned auditory pinnae.

Alterations of both Foxg1 transcriptional dynamics and the structural organization of the mature λjunction in Foxg1 -/-embryos
Both the perturbation of the development and pattern of external SCE derivatives and the significant re-patterning and dysmorphogenesis of the upper jaws, OPC, and NC evinced in Foxg1 -/-mutants suggested that the developmental elaboration of the λ-junction was affected by the loss of Foxg1. To initiate an investigation of possible disruptions in the ontogeny of the λ-junction, we examined the λ-junction of Foxg1 -/embryos and their littermates just after 'maturation' (as seen at E11). This developmental stage occurs just after the establishment of significant relative topographical relationships of substantial and essential patterning information that is manifest around E10.5 (7,9).
We first investigated the patterns of transcriptional activity of the endogenous Foxg1 locus in mutant embryos and compared them to those of their Foxg1 -/-littermates. At E11, the pattern of activity detected in Foxg1 +/-embryos followed that described above; Foxg1 -/-mutant embryos, however, exhibited a distinct pattern of activity within the SCE associated with the λ-junction and the development of the upper jaw ( Figure   3A-F). Unlike their E11 heterozygous littermates, Foxg1 -/-embryos showed more intense, extensive staining in the SCE of the lateral and oral aspects of mxBA1, the trigeminal ganglion, and the optic primordia, and throughout the SCE of the stomodeal roof. A similar aberrant increase was detected in the SCE associated with both the lFNP and mFNP of E11 Foxg1 -/-embryos. Additionally, ectopic X-gal staining was detected at this time point in the epithelium of mutant embryos between the otic vesicle and the epibranchial placodes.
This striking difference in the Foxg1 transcriptional profile between heterozygous and mutant embryos led us to examine the morphology of E11 embryos in greater detail. This assessment highlighted five further areas of morphologic distinction visible by light microscopy between the E11 Foxg1 -/-embryos and their heterozygous littermates ( Figure 3). First, in Foxg1 -/-embryos the FNP and mxBA1 were positionally reoriented and rotated ventrolaterally, an alteration accompanied by an increase in the breadth of Rathke's pouch and the entire stomodeum. Second, this increased stomodeal breadth was associated with a decrease in epithelial thickness of Rathke's pouch but a general increase in the thickness of the epithelial layer of the remainder of the SCE of the stomodeal roof. Third, the olfactory pits of mutant embryos were smaller and less elongate. Fourth, the architecture of the developing lens and optic cup was already clearly anomalous and disorganized with the loss of Foxg1. And fifth, a distinctly anomalous, dorsolateral bulge of tissue protruded from the mxBA1 ventro-caudal to the eye and dorsal of the mxBA1-mdBA1 boundary of mutant embryos.
These trends were still patent at E11.5, though this time point coincided with additional morphological changes of mxBA1 and the lFNP. For instance, rather than a single nasolacrimal groove running from the developing orbit to the center of the λ-junction, E11.5 Foxg1 -/-embryos displayed two additional (ectopic) shallow grooves within mxBA1 and a dorsal reorientation of the groove between mxBA1 and the lFNP.
Moreover, the invaginating olfactory pits of Foxg1 -/-mutants continued to be both less deep and less elongate.
By E13.5, Foxg1 heterozygous embryos exhibited, as expected, β-galactosidase activity along the rim of the invaginated lens vesicle and in the rostral optic cup as well as in the forming nasolacrimal ductal system exiting the developing orbit (Figure 3). High levels of activity were further detected in the epithelia associated with the external nares and in discrete vibrissae primordia. Low levels of activity were also detected along the upper lip and vibrissal pad; however, the maxillary region below the eye was distinctly absent of even low-" 13 level activity. By comparison, E13.5 Foxg1 -/-embryos displayed β-galactosidase activity throughout most of the malformed eye, as well in a large area of surface epithelium associated with the abortive nasolacrimal ductal system leaving the orbit. Additionally, a large swath of lightly positive staining was detected in the maxillary epithelium from the eye and nasolacrimal region to the margin of the upper lip. Thus, the loss of Foxg1 clearly alters both the dynamics of its own transcription and the ontogenetic elaboration of the upper jaw-associated facial primordia morphogenesis.
Structural reorganization at the λ-junction in Foxg1 -/-mutants includes the loss of the coordinated directionality and orientation of the oral epithelium, the loss of internal choanae, and impaired elevation of the palatal shelves.
The observed gross changes in regional morphology, including in the apparent increased thickness of the oral epithelium, coupled with regional changes in the transcriptional activity of the endogenous Foxg1 locus, led us to further examine the embryonic cranial epithelium. Because scanning electron microscopy (SEM) is known to provide crisp, detailed images of surface characteristics of embryos (137,138), we utilized SEM to examine the cranial epithelia of Foxg1 -/-embryos and their heterozygous littermates.
Removal of the caudal, post-maxillary embryo permits visualization of the oral aspect of the λ-junction as well as the stomodeal roof and internasal groove. While low magnification SEM micrographs of E11.5 Foxg1 -/-embryos clearly showed the wild type demarcation of landmarks associated with the λ-junction as well as the stomodeal roof and internasal groove (detailed in Figure 1), micrographs of mutant embryos both confirm defects already noted by light microscopy and reveal less obvious alterations in the orientation and/or topography of these same landmarks ( Figures 4A, B). For example, the internasal line and nasolacrimal groove were seen to be shallower and shorter in mutants embryos, while the ventral intranasal line, stomodeal-palatal line, and oronasal groove were each found to be shallower and longer. Additionally, SEM micrographs of mutant embryos confirmed that mutant embryos were characterized by the re-orientation of the FNP as well as a medio-lateral expansion of the stomodaeum and Rathke's pouch.
SEM micrographs, moreover, revealed further morphological changes in E11.5 Foxg1 -/-embryos undetected by light microscopy (Figures 4 A', B'). Normally, the surface of the stomodeal roof of E11.5 embryos exhibits a fan-like sets of ridges radiating rostro-laterally from Rathke's pouch to the mFNP and to where the internal choanae are beginning to invaginate (137,138). However, E11.5 Foxg1 -/-embryos showed no initiation of choanal invagination and exhibited fewer such ridges; additionally, those ridges that were present were un-associated with Rathke's pouch, and abnormally ran medio-laterally. Moreover, because light microscopy suggested that the oral epithelium was thicker in Foxg1 -/-embryos, we examined it at higher magnification. We found that the usual coordinated cellular directionality, size, and orientation characteristic of the epithelial surface of Foxg1 +/-(and wild type) embryos was lacking in Foxg1 -/-embryos: For instance, along the oral surface of mxBA1 the surface cells typically are of relatively uniform size and the long axis of the cells is nearly universally oriented to extend along the long axis of the arch (roughly parallel to the ridges noted above). This was not found to be the case in E11.5 mutant embryos as the cells of the oral surface of mutant mxBA1 varied widely in size and shape, and many cells of the mutant embryos were otherwise mis-oriented along the short axis of the arch.
Furthermore, in normal murine embryos from E11.5 to E12.5 the primary choanae typically invaginates where the developing primary and secondary palates laterally converge to generate antra on each side of the caudal end of the future nasal septum; eventually, the internal, deep epithelia of each antrum will " 14 make contact with the epithelia of the corresponding ventrocaudal invagination of the nasal cavity thus forming the oronasal (bucconasal) membrane. Rupture of this membrane thus permits complete passage from the external nares to the posterior respiratory system. Choanal atresia, effectively the developmental obstruction of the choanae, was clearly presaged in SEM micrographs of Foxg1 -/-embryos at E12.5 ( Figures   4C, D). Moreover, while both the mutant primary and secondary palatal primordia were present and positioned appropriately they were also both less clearly robust and dysmorphic. The trend toward a medio-laterally expanded stomodeal roof seen in younger mutant embryos was still patent with the broader primary palate and the laterally displaced upper lip primordia evinced in older embryos. Concordant with the abnormalities observed in the upper jaw skeleton, the leading edges of the secondary palate were seen to barely descend from the stomodeal roof in Foxg1 -/-embryos.
Additionally, by E12.5 the murine oral epithelium has normally generated a specialized transient superficial cellular population, the periderm, thought to act as a protective barrier that inhibits potentially pathological adhesions between otherwise immature, adhesion-competent epithelia (139)(140)(141)(142)(143). SEM We found that the loss of Foxg1 fundamentally altered the topography of the expression patterns of these genes at these stages ( Figure 5). For instance, at E9.25 Bmp4 (a member of the large Tgf-β superfamily of signaling molecules) is dynamically expressed in the SCE of Foxg1 +/-embryos, including at the presumptive center of the future λ-junction as well as in the distal mdBA1, the first pharyngeal plate, and epibranchial placodes, and the dorso-caudal optic primordia ( Figures 5A-D) (48-50, 144-151). It is also expressed in cells in the region of the commissural plate (CP) that derives from the ANR. Within the forming λjunction, Bmp4 mRNA expression is typically detected in cells overlying the portion of mxBA1 growing toward the olfactory placode and in a dorsally projecting stripe delimiting optic SCE from olfactory SCE. In Foxg1 -/-" 15 embryos, however, Bmp4 was expressed more broadly in cells of the optic primordia, the forming λ-junction and the epibranchial placodes. Moreover, in Foxg1 -/-mutants Bmp4 expression in the dorsally projecting stripe associated with the λ-junction was stronger and extended further both dorsally and caudally toward the optic primordia. Additionally, precocious Bmp4 expression was seen running through the SCE of the olfactory region and abnormally extensive expression was detected associated with the CP.
Aberrant Bmp4 expression was also seen associated with the mature λ-junction of E10.5 Foxg1 -/embryos. Normally, strong but tightly restricted and focalized Bmp4 expression is found in the epithelia of the distal tip of mxBA1 and along the ventral rims of the invaginating olfactory pits, in a small swath of cells on the lFNP directed toward the eye, and in the odontogenic line paralleling the naso-stomodeal line along the ventral mFNP. At this embryonic stage, lighter Bmp4 mRNA expression is also typically detected in the dorsal rims of the olfactory pit, the cells along the future nasolacrimal groove, and the dorso-caudal optic primordia.
In E10.5 mutant embryos, Bmp4 expression was found to be strong at, but not tightly restricted within, the λjunction. Evidence of greater levels of Bmp4 mRNA in mutant embryos was seen along the dorsal olfactory pits and lower levels seen associated with the ventral mFNP. Noticeably, expression associated with mxBA1 differed greatly between heterozygous and homozygous mutant embryos. For instance, rather than being expressed in cells associated with the nasolacrimal line, ectopic Bmp4 transcripts were detected in a broad swath of SCE cells running ventral to the eye and toward the ectopic, sub-optic maxillary bulge (described above).
Fgf8, a developmentally active and potent fibroblast growth factor, is essential for the normal development of the jaw (152)(153)(154)(155)(156)(157) as well as the OPC (158) and NC (158)(159)(160)(161)(162), and ontogenetic disruption of its expression characterized E9.25 and E10.25 Foxg1 -/-embryos ( Figures 5I-P). In addition to the isthmic organizer, at E9.25 Fgf8 mRNA is normally highly detectable in the oral ectoderm of BA1 and the SCE overlying the developing CP. Notably, Fgf8 is also typically expressed in a more diffuse manner in the ventrolateral ectoderm (VLE), a domain of the SCE that lies between the CP and the forming olfactory placode. At E9.25, Foxg1 -/-embryos exhibit significant loss of detectable Fgf8 mRNA in the VLE just lateral to the CP. Expression in the CP itself was maintained in mutant embryos but did not extend as far dorsoventrally within the CP as in Foxg1 +/-embryos. Moreover, low levels of Fgf8 transcripts were also detected extending further medially from the mxBA1 oral epithelium toward Rathke's pouch in mutant embryos.
An embryonic day later, the Fgf8 expression profile of the λ-junction in Foxg1 -/-mutants continued to be anomalous (Figures 5I-P). In wild type embryos at E10.25 Fgf8 mRNA is clearly detected in the isthmic organizer, dorsal diencephalon, and the ventral optic stalk. It is also found in the oral epithelium of mdBA1 and mxBA1 as well as in the outer-margins of the dorsal epithelium of both mFNP and lFNP that rims of the olfactory pit; at this stage, however, Fgf8 is normally distinctly absent from the center of the λ-junction where mxBA1 and the FNP converge. Transcripts of Fgf8 also continue to be expressed in a diffuse manner in the VLE of the mFNP. In mutant embryos, however significant Fgf8 mRNA expression was found abnormally associated with the center of the λ-junction and cells rimming the entire olfactory pit. Within the mutant pits, Fgf8 was aberrantly expressed in the medial wall where the primordial cells of the VNO are thought to arise.
Foxg1 -/-mutants also failed to express Fgf8 in the VLE although they did ectopically express it rostrally and dorsally throughout the NE from the malformed ventral optic stalk to the dorsal diencephalon.
Retinoic acid (RA), another potent signaling and patterning molecule, is a lipid soluble morphogen generated as an active derivative of vitamin A and its developmental regulation must be tightly controlled (148)(149)(150)(163)(164)(165)(166)(167)(168). Raldh3 is a gene encoding a retinaldehyde dehydrogenase involved in the second oxidative step in RA synthesis, and it is focally expressed at the heart of the developing λ-junction in murine " 16 embryos from E9 to E10.5 ( Figures 5Q-V). At E9.0, Raldh3 mRNA is normally found within the SCE associated with the optic primordia, the distal mxBA1, and the ventral olfactory pit of both the mFNP and lFNP.
By E10.5, Raldh3 is conspicuously, though asymmetrically, expressed along the rim of the invaginated lens and optic cup as well as in the epithelium of the ventral-most invaginated olfactory pit. However, in E9.0 Foxg1 -/-embryos Raldh3 expression was not as extensive in the SCE associated with the optic primordia or mxBA1. By E10.5, a significant reduction in olfactory Raldh3 expression was visible in mutant embryos.
Moreover, while expression associated with the dysmorphic eye was more intense, including in the cells of the lens and optic cup, the asymmetric nature of Raldh3 expression was maintained in mutant embryos.
Shh encodes another secreted protein, one with autoproteolytic activity that, in conjunction with covalent addition of cholesterol, produces both long and short-range morphogens (169)(170)(171). It is embryonically expressed in many ventral, midline tissues including the node, head process, notochord, and floor plate and basal ganglia of ventral forebrain, and its loss leads to dramatic holoprosencephaly (150,(172)(173)(174)(175)(176)(177). It has been previously reported that, because Foxg1 mutants lack a proper ventral forebrain, they also lack Shh in this region (105). We confirmed the absence of Shh in the ventral forebrains of E10.

Embryos
In addition to epithelial signals, the mature λ-junction at E10.5 is a source of signaling molecules originating from the CNC ectomesenchyme. This includes production of Wnt5a, a protein that signals through non-canonical Wnt pathways (178)(179)(180)(181)(182). Typically, Wnt5a expression at E10.5 is strong within the mFNP, the lFNP and the portion of mxBA1 associated with the FNP at the center of the λ-junction, while weaker expression is encountered in the distal mdBA1 ( Figure 6). In Foxg1 -/-embryos, Wnt5a mRNA was detected in the λ-junction; however, Wnt5a expression was weaker than normal in mxBA1 but stronger than normal throughout the FNP and olfactory pit. Moreover, a noticeable gap in expression between the FNP and mxBA1 was encountered in Foxg1 -/-mutants. Thus, we found the topographic elaboration of essential craniofacial patterning genes in both the embryonic cephalic ectoderm and mesenchyme to be significantly altered in Foxg1 -/-embryos from E9 through E10.5.
Because, in the normal course of epithelial-mesenchymal cross-talk, cell and tissue competence to signal is typically coordinated with cell and tissue competence to respond, we examined the cephalic gene expression patterns of early response and obligatory downstream effectors of signaling and compared the expression topographies of presumed molecular signal and response. Herein we highlight alterations detected in Foxg1 -/-embryos of the expression in two such genes: Spry1, an early Fgf8-responsive gene and mediator of Fgf8 signaling (183), and Ptc1, an effector of Hedgehog signaling (184). As one might expect for an early responsive gene, Spry1 expression in E10.5 murine embryos generally follows that of Fgf8: thus, in Foxg1 heterozygotes it was expressed at the isthmic organizer, the dorsal diencephalon, the CP, the BA1 oral ectoderm, the pharyngeal plates and rimming the entirety of the olfactory pit and within the nasal fin ( Figure   6E, G). Though we did not discriminate this in our whole mount in situ hybridization assays, low levels of Spry1 have also been reported to be expressed in the lens and corneal epithelium where it is necessary for proper separation of the lens vesicle from the SCE and the formation of eyelids (185,186). We found that in most tissues Spry1 expression in E10.5 Foxg1 -/-embryos followed that of Fgf8 expression ( Figures 6F, H). For " 17 instance, mutant embryos expressed Spry1 in the oral epithelium and in an expanded manner associated with the expanded Fgf8 expression in the dysmorphic rostral brain (including the optic sulcus). Notably, however, detection of Spry1 mRNA in the mutant FNP was restricted to low-levels at the ventral end of the olfactory pit and did not mirror the strong ectopic expression of Fgf8 at the center of the λ-junction and along the olfactory pit. Thus, unlike in other embryonic regions the topography of Fgf8 expression associated with the λ-junction was not followed by similar topography in the expression of this Fgf8 responsive gene. Moreover, as with Fgf8 and Spry1, Ptc1 mRNA is typically expressed in the olfactory pits associated with mature λ-junctions ( Figure   6). Ptc1 acts, in addition to being as an effector of Hedgehog signaling through its role as a cell-surface receptor, as an early responsive gene to Shh signaling (184). We found its expression in the olfactory pit at E10.5 abrogated in embryos lacking functional Foxg1.
The evident ontogenetic perturbations of signaling gene expression exhibited in Foxg1 -/-embryos prompted us to further investigate the molecular elaboration of the mature λ-junction in mutant embryos. We proceeded by characterizing the patterns of expression of genes and proteins implicated in disparate aspects of normal, regionally relevant cranial development, including Sox10, Pitx2, Eya2, Six3, and endomucin (an early marker of forming blood vessels).
Sox10 is expressed in CNC as they delaminate and then migrate from the neural tube-ectodermal boundary (187)(188)(189)(190)(191). By E10.5, the CNC have already ceased their migration and are executing their responses to local developmental cues; at this point, Sox10 mRNA is found associated with the cranial nerve ganglia and the presumptive olfactory ensheathing cells (OEC) within the mesenchyme of the olfactory pit (188,(192)(193)(194). The cranial ganglia are composite structures formed of both CNC and cells that have delaminated from the particular neurogenic placode; this process of gangliogenesis requires tightly controlled epithelial-CNC interactions and proper expression of Sox10 during this process occurs in the context of a normally developed epithelial-mesenchymal interface. SOX10 has also been implicated in Kallman syndrome wherein a disruption occurs in the establishment of the routing of neuroendocrine gonadotropin-releasing hormone (GnRH) cells along the vomeronasal and terminal nerve fibers in the peripheral olfactory system to their final destination in the preoptic and hypothalamic region of the forebrain (195). It is also expressed in the otic vesicle (196, 197; Figure 6K). In E10.5 Foxg1 -/-embryos we found that (where expressed) Sox10 mRNA was strongly detectable; however, the pattern of Sox10 expression was topographically altered ( Figure 6L).
For instance, we failed to detect Sox10 mRNA in the FNP. Staining within the trigeminal ganglia, moreover, revealed both an abnormal shape to this structure and abnormality of ophthalmic branching and extension Loss of Six3 (a sine oculis-related homeobox transcription factor gene) in mice results in significant tissue loss in the forebrain as well as structures associated with the SCE and λ-junction (212)(213)(214)(215)(216)(217). Six3 heterozygosity has also been implicated in anosmia-based decreased male fertility due to disruption of GnRH cells migration stemming from defective olfactory axon targeting (218). Because development of the nascent embryonic vascular plexus and its remodeling is believed to be sensitive to the local molecular environment (219-221), we assayed cephalic vascular development in E10.25 Foxg1 +/-and Foxg1 -/-embryos through whole mount anti-endomucin immunostaining. This revealed that while vasculogenesis and subsequent angiogenesis were not globally impaired in mutant embryos, there were regional changes in the cephalic vascular network ( Figures 6Y-β). For instance, the mutant otic plexus was underdeveloped, with an expanded vascular-free zone, and was closely apposed to the dorsally-shifted anterior cardinal vein ( Figure 6Z). The optic plexus was both expanded and disorganized, while the trigeminal plexus was smaller and constricted. Moreover, the vascularization of the FNP rimming the olfactory pit was more chaotic, the typically tight capillary sheet covering the telencephalon failed to manifest as such, and a larger region of the midline stomodeal roof remained largely vessel free in Foxg1 -/-embryos.

Loss of Foxg1 Significantly Alters SCE Compartmentalization and The Topography of Apoptosis
Because our previous analysis had demonstrated (1) that the cellular organization of the SCE in However, when we investigated the cellular organization and appearance in the normally Foxg1 + olfactory-associated SCE rostral to the eye at E9.25, we noted differences between Foxg1 +/-and Foxg1 -/embryos. For instance, the Foxg1 +/-SCE exhibited much rounder (less flattened) cells interspersed with numerous cellular extrusions typically indicative of apoptotic cell death (compare the yellow bordered insets in Figure 7F and G). Additionally, examination of scanning electron micrographs of the normally Foxg1-positve SCE associated with the hinge region between mxBA1 and mdBA1 showed that both the Foxg1 +/-and the Foxg1 -/-embryos exhibited rounded cells; notably, the SCE of the mutant embryos in this region also exhibited numerous apparently apoptotic cellular extrusions (compare the red bordered insets in Figure 7F and G).
" 20 Because scanning electron micrographs suggested distinct topographies of putatively apoptotic cellular extrusions within the Foxg1-positive SCE, we utilized whole-mount TUNEL assays of E9.25 Foxg1 heterozygous and mutant embryos to confirm the topography of apoptotic cell death. Supportive of the scanning electron data, we found that the normally Foxg1-positive hinge-centered oral and caudal (i.e., aboral, peri-cleftal) mxBA1 SCE associated with the hinge presented an intense clustering of apoptotic cells in cleft. Epibranchial and otic associated mutant ectoderm also displayed greater numbers of apoptotic cells.
Notably, a swath of apoptotic cells in mutant embryos did not entirely rim the lens pit but were only found in the rostral and dorsal compartments, the latter of which formed the base of a continuous patch of apoptotic cells dorsally extending to a highly apoptotic supra-orbital region. Moreover, apoptotic cells ran caudally as a continuous sheet from the trigeminal region, over the lens, and through the groove between the telencephalic and frontonasal protrusions. The ectopic pattern of cell death encountered over the middle of the telencephalon at E9.25 was also clearly patent at E10.75. Rather than running toward the eye, cell death associated with nasolacrimal groove was re-oriented toward the ectopic mxBA1 bulge (which itself remained relatively free of cell death; yellow arrowheads, Figure 7M) in

Loss of Foxg1 leads to the ectopic acquisition of deep mandibular (lower jaw) molecular identity in the maxillary (upper jaw) primordia
Extensive experimental evidence has demonstrated that two key features of gnathostome jaw bulge. Like Dlx2 (though to a lesser extent), expression of Dlx3 was aborgated within the λ-junction eithelium, while expression was also detected (like Dlx5) within the core of the ectopic mxBA1 bulge ( Figures 8M, N).
To confirm that the CNC of the ectopic mxBA1 bulge had acquired a mdBA1 molecular identity outside of de-regulated Dlx expression, we examined Pitx1 expression at E10.5 and E10.75 in Foxg1 heterozygous and null embryos. Normally, Pitx1 is expressed within a distal subdomain of mdBA1 CNC and not in the mxBA1 CNC topographically akin to the ectopic bulge ( Figures 8O, Q). However, in the absence of  Figure 8Z, β) while expression at the center of the λ-junction and the mxBA1 oral epithelium was discontinuous and disorganized. Furthermore, rather than discretely lining the nasolacrimal groove, mutant embryos showed a broad, caudally shifted patch of elevated p63 expression; notably, this patch of elevated p63 expression was discretely offset by a patent decrease in the level of p63 associated with the ectoderm of the ectopic bulge in mxBA1. Moreover, we found an altered topographic profile in p63 signal in the cells of the lens pit of Foxg1 -/-embryos.

Foxg1 genetically interacts with Dlx5 in cranial skeletal development and SCE organization
Together, our results have demonstrated that Foxg1 regulates considerable aspects of non-NE " 23 craniofacial development. To approach a greater understanding of the broader context in which Foxg1 regulates SCE development and craniofacial patterning, we began a systematic examination of potential genetic interactions of Foxg1 and other regulators of λ-junction associated SCE and cranial skeletal development. Among these regulators is Dlx5, which, in addition to being expressed in the CNC of mdBA1, is extensively expressed in the early embryonic SCE before being restricted to the olfactory pit around E10.5.
We present here direct evidence of a significant genetic interaction between Foxg1 and Dlx5 for SCE development and cranial skeletal patterning.
Elucidation of a genetic interaction between two genes involves a demonstration that the progressive loss of individual alleles of the two genes results in a distinct (not simply additive) phenotypic alteration, and examination of the phenotypic consequences of the progressive loss of Dlx5 alleles in a Foxg1-null background clearly demonstrates that, with regard to cranial development, Dlx5 and Foxg1 genetically interact ( Figure 9). The phenotypic alterations evinced in Foxg1 -/-;Dlx5 +/-and Foxg1 -/-;Dlx5 -/-embryos and perinates are extensive, and therefore our descriptions herein are not exhaustive and are limited to those aspects pertinent (1) to the demonstration of a genetic interaction and (2) to the elucidation of the nature of this interaction.
We found that compound heterozygous Foxg1 +/-;Dlx5 +/-mice were viable and fertile, and when interbreed they generated the expected ratios of wild type, single heterozygous, compound heterozygous, heterozygous plus homozygous, and both single and double homozygous genotypes, including Figure 9). Foxg1 -/-;Dlx5 +/+ perinates from Compound Foxg1 -/-;Dlx5 +/-mutants were easily distinguishable from Foxg1 -/-mutants ( Figure 9); their snouts were noticeably shortened rostro-caudally and, while a reduced rhinal sulcus medianus was patent, the remainder of the rhinarium and external nares were not evident (Fig 9D3). While a rostroventral neurocranial midline was represented in compound mutants by a TBP it was more hypoplastic than that of the single Foxg1 -/-mutants but was rostrally continuous with a cylindrical, rostro-caudally compressed rod-like nasal septum. Notably, nasal capsules and a true cavum nasi were absent (i.e., there were no pars anterior, pars intermediale, and pars posterior); rather, three dysmorphic cartilaginous structures were present associated with the rostral neurocranial midline. The first was a dorso-ventrally oriented, discontinuous plate (shaded lighter green in Figure 9D5) and the other two were bilateral laminar structures perpendicular to the first and likely representing the remnants of the paired lamina transversalis posterior (shaded darker green in Figure 9D5). These NC defects occurred concomitant with defects of the regional dermatocranial bones, including the premaxillae, maxillae, and palatines, and are associated with development at the λ-junction.
Moreover, the TT of compound mutants was also found to be detached from the otic capsule but was synchondrotic with the re-oriented styloid process (stylohyal plus tympanohyal); together, this fused 'stylotegmenal' structure was widely separated from the incus (unlike in the Foxg1 -/-mutants), which was itself " 24 further dysmorphic with its crus longus re-oriented toward the stylo-tegmen. The loss of a single Dlx5 allele in a Foxg1 null background, moreover, resulted in further hypoplasia of the pars canalicularis of the otic capsule.
Importantly, each of these defects of the craniofacial skeleton and associated soft tissues are sufficient to demonstrate the genetic interaction of Foxg1 and Dlx5. Each rod was bounded on either side by dermatocranial bone that was, in turn, associated with a dysmorphic, diminished incisor. These four rod-associated ossifications extended caudad to highly hypoplastic presumptive premaxillary bodies that were synostotic with likewise highly dysmorphic maxillae. The entire palatal space was found to be chaotic and cleft, not elevated, and characterized by numerous isolated palatine ossifications. Additionally, the elements of the middle ear and the otic capsule evinced further degradation of structures (data not shown). Notably absent, moreover, was the straight rod-like post-optic pillar seen in Foxg1 -/-and the Foxg1 -/-;Dlx5 +/-mutants ( Figure 9E5).

Fundamental re-patterning of the SCE and LJ of Foxg1 -/-;Dlx5 -/-embryo.
Because the Foxg1 -/-;Dlx5 -/-double mutants showed such a drastic re-patterning of their anterior cranial skeleton, we posited that E10-E11 double mutant embryos would likewise evince fundamental architectural changes correlated with reorganized patterns of gene expression within the cephalic ectoderm.
Indeed, when examined at E10.25, in addition to the hypoplastic telencephalon vesicles seen with the loss of Foxg1, the Foxg1 -/-;Dlx5 -/-double mutant embryos were found to be characterized by an absence of an olfactory pit (though there appeared to be some subjacent mesenchyme) and the truncation of the FNPassociated mxBA1 ( Figure 10B, D). Notably, compound mutants did not exhibit the ectopic budge evident in the single Foxg1 -/-mutants.
Both of these aberrant phenotypes are plausibly correlated with molecular changes at the λ-junction.
Therefore, as an initial read-out of the patterning road-map of the cephalic ectoderm and the λ-junction we examined the expression patterns of Bmp4 at E10. 25 and Fgf8 at E10.75 ( Figure 10). While we found, as expected, that normal distal mdBA1 Bmp4 expression was robust in double mutant embryos, we also unexpectedly found a near complete loss of junctional Bmp4 expression. Loss also characterized the first pharyngeal plate, the epibranchial placodes and the otic primordia of Foxg1 -/-;Dlx5 -/-double mutant embryos.
The region of the antero-ventral CP, however, showed increased expression, as did (though to a lesser " 25 degree) the external optic primordia.
While the ectoderm of the double mutant was incapable of invaginating to form either a lens pit or an olfactory pit, the presence at E10.25 of frontonasal mesenchyme that subsequently yielded three structural midlines suggested that positional patterning was being imparted to the ectomesenchyme. As dynamic Fgf8 expression is known to be essential for NC development and polarity as well as the establishment of the midline (158), we examined its expression at E10.75. We found that singular foci of epithelial  Foxg1 is a significant craniogenic regulator of both the 'contents' and the 'container'. Craniogenesis, the developmental process of elaborating a fully functional, integrated head, requires the coordination of developmental mechanisms regulating both the formation of the brain (i.e., the 'contents') and the skull (i.e., the 'container'). Normal development of the contents and the container each require mechanisms that yield complex structures from precisely established developmental fields; and such developmental mechanisms for one must be, at some level, attuned to those of the other -a notion that has been clinically canonized in the aphorism that 'the face predicts the brain' (236). Indeed, coordinated cranial integration is reflected in the great numbers of both murine mutants and human patients suffering from both craniofacial malformations and coexisting brain malformations. Evidence presented herein, in combination with previous work on neurodevelopment, makes clear that Foxg1 acts as a significant craniogenic regulator of both the contents and the container. . CRC are regionally induced by the three dorsal, regional signaling centers (CP/pallial septum, the CH, and the anti-hem), are highly motile, and are thought to be the earliest post-mitotic cortical neurons to develop (246,247). In mice, CRC normally exit the cell cycle beginning at E10.5 and migrate tangentially to evenly cover, through a combination of random walk migration and inter-CRC contact repulsion, the preplate/margial zone of the entire embryonic neocortex and, in a foundational manner, create an initial cortical layer 1. There, CRC interact at the neuroepithelial basement membrane with the nascent pial meninges and with the end feet of the pseudostratified neuroepithelia to regulate, principally through their expression of Reelin, the sequent layered, radial migration of pyramidal cells (248)(249)(250)(251)(252)(253)(254)(255)(256)(257). This is a fundamental stage in corticogenesis but one that must be halted once completed. Foxg1 acts to suppress CRC fate and thus expression is normally excluded from neuroepithelial cells in CRC producing regions (in a mechanism likely involving direct targeting by miR-9 (123)). Foxg1 expression is strong, however, in the remainder of the necessarily mitotically active neuroepithelium where it acts to maintain the neuroprogenitor pool necessary to layer the cortex (110,111). It is subsequently tied during cortigenesis to the regulation of the necessary temporal changes in competence to generate upper layer and deep layer neocortical neurons.
Notably, Foxg1 expression is transiently downregulated as pyramidal neurons migrate through the neocortical intermediate zone and initially adopt a multipolar shape (115). Once the cells continue to their " 29 specific destinations in the cortical layers Foxg1 is again up-regulated. Cells that fail to transiently inhibit Foxg1 expression during this period stall their migration and subsequently fail to find their proper cortical layer.
Pyramidal neuron precursors in their multipolar migratory phase and CRC share in common their predilection for decreased cell-cell connectivity, tangential migration, expression of Reelin, and repression of Foxg1

expression. Moreover, telencephalic GABAergic interneuron precursors likewise selectively downregulate
Foxg1 during the tangential phase of their migration and reinitiate expression when they have invaded the cortical plate, suggesting that tangential migration in the forebrain universally requires transient repression of Foxg1 (115). These observations also suggest that Foxg1 regulates epithelial cell-cell interactions and cohesion as expression is lost when intra-epithelial cohesion is lost and subsequent migration ensues.  Thus, the disruption of midline patterning of the presphenoidal TBP appears mirrored by disruption in the overlying midline optic chiasm.
Moreover, the loss of OPC and related structures in Foxg1 -/-skulls is particularly striking when viewed from norma lateralis (compare Figure 2E, F) where a complete absence of orbital skeletal structures permits visualization through the entire breadth of a cleared mutant neonatal skull, indicating that the cavum cranii here is utterly unprotected. This striking detail is not likely to be due simply to the loss of optic tissue. Indeed, orbital skeletal structures are present and patterned in the absence of retinal formation (e.g., Rx -/-skulls; 282); and, in fact, the absence of an entire eye, such as with Pax6 Sey/Sey mutants, does not preclude the formation of OPC (or even an optic foramen; 78) or frontal bone orbital laminae. It is more likely that the lack of skeletal structure in Foxg1 -/-mutants results from a knock-on effect due to the chaotic expansion of retinal tissue along with loss of appropriate skeletal patterning signals to the perioccular CNC.
The position and aberrant shape of the Foxg1 -/-post-optic pilar is notable for a number of reasons.
Second, this bulge is filled with Dlx5 + ectomesenchyme. And third, Dlx5 is known to regulate both the straight, rod-like cylindrical shape and the cellular differentiation of MC in the lower jaw (6, 7, 79). Whether there is a direct correlation between the Dlx5 + ectopic bulge and the aberrant Foxg1 -/-post-optic pillar remains unclear, " 32 consideration must be given to the fate of these Dlx5 + cells. Notably, Foxg1 -/-;Dlx5 -/-double mutant embryos lack -in addition to the regulatory effects of Dlx5 -both an ectopic bulge and post-optic pillars of any kind.
Development in Foxg1 -/-mutants of the vestibular and cochlear canals of the OTC has received some previous attention (104,118). In addition to hypotrophy of both the pars canalicularis and pars cochlearis housing the canals (as previously described), we found that the oval and round windows of mutants appeared anomalous in size and orientation. The mammalian OTC is also normally perinatally synchondrotic with a number of extra-capsular structures, including the TT, the styloid process, and the alicochlear commissure; we found each of these structures to be dysmorphic and/or heterotopic in Foxg1 -/-perinates. However, because each of these structures are also associated with the viscerocranium we discuss them in this context below.
Dermatocranial elements also make significant contributions to the calvarial neurocranium. These  " 34

Foxg1 expression, the SCE, and epithelial morphogenesis.
We initiated our investigation, in part, because Foxg1 is expressed in what we posited were the right places and at the right times to play a key role in integrating development of both the content and the container during craniogenesis. Therefore, we examined embryonic Foxg1 expression in the nascent and maturing SCE in more detail. Our β-galactosidase activity and confirmatory in situ hybridization analysis, taken together with that of previous investigations (e. g., 103, 130, 241), highlights several perhaps underappreciated points regarding Foxg1 expression and cranial development. The model also ensures developmental tractability in responding to evolutionary changes in functional demands on jaws by positing that the 'hinge' signaling is integrated with 'caps' patterning signals emanating from the distal mdBA1 ectoderm and the SCE at the proximal, mxBA1 end of BA1 (i.e., the λ-junction), two epithelia that share a very early developmental history (8,9). Thus, the baseline of patterning jaws would have fungible CNC interpreting in mirror image fashion Hinge and Caps epithelial signals, and coordinated capsrelated patterns of gene expression would be predicted by the model. Such patterns are indeed conserved in sharks, chicks and mice (322). Developmental modularity and integration are key mechanisms bridging development and evolution, and inherent within the patterning mechanism suggested by the Hinge and Caps model is the potential for developmental modularity within jaw units, a notion likewise supported by experimental evidence (30, 31). Ultimately, however, the model makes clear that understanding jaw development follows from further understanding the origins of, and influences on, the epithelial patterning centers. Notably, the model considers two additional factors. First, the upper jaws develop in more intimate association with the developing brain anlage, and thus upper jaw CNC has to contend with the interpretation of secreted signals otherwise being utilized to pattern the brain, while the lower jaw CNC has to contend with signals regulating the embryonic, pharyngeal vascular entry to the heart. Second, gnathostomes outside of chondrichthyans evolved a premaxillary component, derived from the FNP, in their upper jaw arcades and hence evolved a patterning organization to suit. Thus, the model predicts that there will be, on top of a " 37 common, mirror-image baseline of gene expression associated with BA1, factors independently influencing upper and lower jaw primordia. As with the 'Hinge'-centered and 'Caps' coordinated patterns of gene expression noted above, the predicted presence of gene expression pattens unique only to upper jaw (e.g.,

Raldh3
) or to lower jaw primordia (e.g., dHAND) is encountered and conserved in gnathostome embryos from sharks to amniotes (322). Importantly, the model posits that the λ-junction evolved as the coordinating 'Caps' patterning center in organisms incorporating premaxillary components in their upper jaws.
We and others have begun examining the developmental origins, influences, and functions of the λjunctional epithelia (7-9, 68, 158). Here, we add Foxg1 to the list of factors integral to upper jaw development and the establishment of a functional λ-junction. In the absence of Foxg1, the structural organization, orientation and size of λ-junction are altered, and the normal ontogenic topography of encoding signaling molecules (e.g., Bmp4, Fgf8) and transcription factors (e.g., Dlx2/3/5, Sox10, Pitx1/2, Eya2, Six1/3, Msx1, Pax3, p63) crucial to regional development is vitiated. An extensive body of evidence has demonstrated that the epithelia of the mature λ-junction (i.e., around E10.5 in mice) displays complex, refined patterns of morphogenic signals, including (among others) of Fgfs, Bmps/Tgfβs, Wnts, Shh, and RA synthesizing proteins, which regulate the morphogenesis of the craniofacial skeleton and its associated soft tissues (7-9, " 38 Investigation of the Six1 enhancer region has found, moreover, that it is bound by Dlx5 protein and that it also has a Fox transcription factor binding domain which helps drive Six1 expression in the SCE (343). Here we have shown that the topography of expression in the SCE of both Six1 and Dlx5 in E9.25 Foxg1 -/-embryos is altered, in particular in the dorso-ventral condensing of the olfactory placode field and its aberrant extension under the lens and toward the mxBA1. Together with the vitiated Bmp4 and Fgf8 expression patterns, changes in Six1 and Dlx5 in the mutant SCE suggests that, while not required for placode induction, Foxg1 is required for the normal topographic circumscription and compartmentalization of the placodal domains in the SCE. We posit that such circumscription of SCE domains is normally required for the appropriate elaboration of the λ-junction to effectively function as an essential integrating center of complex olfactory, optic, and upper jaw patterning.
The changes in the apoptotic profile of the SCE are particularly relevant in this light. As a normal, regulated cellular process, programmed cell death through apoptosis is well documented and clearly plays a role in organogenesis and tissue remodeling such as occurs during epithelial invagination (344)(345)(346)(347).
Apoptosis is also key to regulating cell number and the removal of transient tissues when and where no longer needed. Within the SCE apoptosis has been observed, for instance, separating the posterior placodes (epibranchial, otic, and trigeminal) (348,349); it has furthermore been detected in the sculpting phases of olfactory pit and cavity elaboration (350) and in the postnatal pruning of olfactory receptor neuron numbers (351). Apoptosis is, moreover, a critical factor in the control of SCE-related ectodermal fusion at points of epithelial-epithelial contact, including closing of the neural tube, the lens and otic vesicles, and the secondary palate, as well as being required in the normal propagation mid facial epithelial seams between the mxBA1, lFNP and mFNP at the λ-junction, a process the dysfunction of which underlies clefting of the lip (68, 303,
With regard to the latter, sufficient evidence of neurodevelopmental deficits stemming from FOXG1 haploinsufficiency or duplication in humans has led to the recognition of 'FOXG1-related disorders' as a distinct clinical classification variously involving microcephaly, mental impairment, autism spectrum disorders, and a congenital variant of Rett (RTT) syndrome (FOXG1 syndrome (OMIM 613454); 355-369). RTT, the typical form of which has been liked to mutations involving MECP2, is one of the most common causes of intellectual disability in girls and is characterized by neurodevelopmental regression after the first year, including of acquired spoken language and hand skills, together with the onset of stereotypic hand movements and gait abnormalities. The FOXG1-related congenital variant of RTT appears earlier in development, usually by six months, and clinically presents a core set of neurodevelopment features, including microcephaly, developmental delay, severe cognitive disability, absence or minimal language development, early-onset dyskinesia and hyperkinetic movement, epilepsy, and cerebral malformation.
Additionally, FOXG1 syndrome is typified by characteristic craniofacial features, including: a prominent metopic suture, large ears, bilateral epicanthic folds, a bulbous nasal tip, depressed nasal bridge, tented or thickened upper lip, prognathism, hypermetropia, and synophyrys (360). A significant linkage between the FOXG1 locus at 14q12 and nonsyndromic orofacial clefting has also recently been reported (370). Notably, we have shown here that Foxg1 -/-mutant cranial dysmorphologies align remarkably with this gestalt of defects, making Foxg1 mutants ideal models to further investigate both the neurological and non-neurological aspects of the congenital variant of RTT, specifically, and craniogenetic development and evolution, generally.

Materials and Methods
Mice. Generation and genotyping of both the Foxg1 (formerly Brain Factor-1) and Dlx5 mice has previously been reported (130). Unless otherwise noted, multiple embryos or perinates were used for each experimental parameter.
Detection of β-galactosidase activity. The Foxg1 mutant alleles (or Foxg1 LacZ/+ ) possess an in-frame βgalactosidase cassette that permits the detection of transcription of the endogenous Foxg1 locus (130). The following day the samples were rinsed 3 times and then washed (6 x 1 hour) in MABT + 0.5 mg/ml levamisole at room temperature and then washed overnight in the same solution at 4º C. The following morning, the samples were again washed in fresh MABT + 0.5 mg/ml levamisole for 1 hour, then twice in NTMT + 0.5 mg/ml levamisole for 10 minutes before incubating in NTMT + 4.5 µl/ml NBT + 3.5 µl/ml BCIP + 0.5 mg/ml levamisole (protected from light) in order to allow the color reaction to proceed. Once suitably developed, the reaction was halted by washing 3 x 5 minutes in PBT. Samples were then photographed and stored in 4% PFA at 4° C.
Scanning electron microscopy. Preparation of embryonic material for electron microscopy followed protocols (79                  " 75